It's been compared with the Wright brothers' quest for flight and the Apollo moon mission. And now that the DARPA Grand Challenge has been met, it holds similar promise for sensor-rich autonomous robotics. "When the Wright brothers flew their little plane, they proved it could be done," said Dr. Tony Tether, director of the Defense Advanced Research Projects Agency (DARPA), after the first three vehicles crossed the finish line on October 8, 2005. "And just as aviation 'took off' after those achievements, so will the very exciting and promising robotics technologies displayed here today."
Two more robots had completed the course by the following day—an exciting gain from last year. When the first Grand Challenge was run in March 2004, the best contestant completed only 7.4 miles (about 5%) of the course; this time, 22 of the 23 finalists surpassed that distance.
Barbara G. Goode
DARPA projects involve high risk and promise dramatic capability advances for the Department of Defense. The agency conceived the Grand Challenge to help meet a congressional mandate to supply the armed forces with autonomous vehicles by 2015. It's said that one third of ground forces aim to be self guided by then. To that end, the basic rules were:
- 1. Travel autonomously on the ground and complete the course in less than 10 hours.
- 2. Stay within boundaries defined by a data file given to teams just before the start.
- 3. Use of GPS and other public signals is allowed, but control commands may not be sent to the vehicle while en route.
When the finalist teams convened at Primm, Nevada (the start and finish point), in the pre-dawn hours they knew nothing of route specifics. At 3:30 a.m. each received a CD containing details; at 6:30 a.m. they set out on the 131.2 mile Mojave Desert course.
Stanley, the 2005 Grand Challenge winner
Six hours, 53 minutes, and 58 seconds later the Stanford Racing Team's Volkswagen Touareg, Stanley, claimed the $2 million prize. Carnegie-Mellon's two red Hummers followed directly.
Novelist Megan Edwards observed from Primm, ". . . team members huddled around the plasma screens, especially when the first robot reached the treacherous Beer Bottle Pass. As I watched Stanley disappear out of camera view, I couldn't help thinking about Apollo spaceships disappearing behind the moon. Would the bot reemerge, or would it plunge permanently out of sight into a rocky gorge? When Stanley chugged around the corner back into view, even the most jaded old reporters cheered right along with all the geeks."
Spider, Cornell's student-devised entry
Strether Smith, who reported on the 2004 Grand Challenge for Sensors, tracked the makeup of teams touting school names. "As far as I could tell there was no student participation in the Stanford car," says Smith. "By contrast, the Carnegie Mellon teams had significant student involvement, and Cornell's team was composed of all students except for the faculty advisor. The latter didn't finish, but I bet the kids learned more."
And learning is key. "The fact that so many vehicles actually met the challenge this year is a demonstration of the incredible engineering talent we have in the U.S.," notes MobileRobots (www.mobilerobots.com) CEO Jeanne Dietsch. "Advances in technologies like those exemplified in the Challenge turn into the new industries that create jobs for the rest of America."
Let's just say that the 2005 Grand Challenge had more than one winner—not the least of which is sensor technology.